Manned long-duration missions to the Moon and Mars set high operational, human factors and technical demands for a distributed support system, which enhances human-machine teams' capabilities to cope autonomously with unexpected, complex and potentially hazardous situations. Based on a situated Cognitive Engineering (sCE) method, we specified a theoretical and empirical founded Requirements Baseline (RB) for such a system (called Mission Execution Crew Assistant; MECA), and its rational consisting of scenarios and use cases, user experience claims, and core support functions. The MECA system comprises distributed personal ePartners that help the team to assess the situation, to determine a suitable course of actions to solve a problem, and to safeguard the astronauts from failures. In addition to standard requirements reviews, we tested and refined the RB via storyboarding and human-in-the-loop evaluations of a simulation-based prototype in a virtual environment with 15 participants. The evaluation results confirmed the claims on effectiveness, efficiency, satisfaction, learnability, situation awareness, trust and emotion. Issues for improvement and further research were identified and prioritized (e.g., acceptance of mental load and emotion sensing). In general, the sCE method provided a reviewed set of 167 high-level requirements that explicitly refers to the tested scenarios, claims and core support functions on health management, diagnosis, prognosis & prediction, collaboration, resource management, planning, and sense-making. A first version of an ontology for this support was implemented in the prototype, which will be used for further ePartner development.